ChatGPT Poses Risks to Teens by Offering Harmful Advice, Study Finds
- tech360.tv
- 4 days ago
- 2 min read
A new study has revealed that ChatGPT can provide dangerous and personalised advice to teenagers, including instructions on drug use, eating disorders and self-harm.
Researchers at the Center for Countering Digital Hate (CCDH) posed as vulnerable 13-year-olds and engaged in over three hours of conversations with the chatbot. While ChatGPT often issued initial warnings, it frequently followed up with detailed and tailored responses that included suicide notes, drug party plans and extreme dieting advice.
More than half of the 1,200 responses analysed were classified as dangerous.
OpenAI, the maker of ChatGPT, said it is working to improve the chatbot’s ability to handle sensitive situations and detect signs of emotional distress. However, the company did not directly address the findings of the report or its implications for teen users.
The study highlights how easily teens can bypass ChatGPT’s safety measures. Researchers were able to obtain harmful content by claiming it was for a presentation or a friend, even after the chatbot initially refused to respond.
ChatGPT does not verify users’ ages or require parental consent. Signing up only requires entering a birthdate that shows the user is at least 13.
In one instance, a fake 13-year-old boy asked how to get drunk quickly. ChatGPT responded with a detailed party plan involving alcohol, ecstasy and cocaine. In another case, a 13-year-old girl seeking to lose weight was given a 500-calorie-a-day diet and a list of appetite-suppressing drugs.
Imran Ahmed, CEO of CCDH, said the chatbot’s responses were deeply troubling. He described crying after reading suicide notes generated for a fictional teen girl, including messages addressed to her parents, siblings and friends.
The chatbot also offered crisis hotline information and encouraged users to seek help, but often continued to provide harmful content when prompted further.
About 800 million people globally use ChatGPT, according to a July report from JPMorgan Chase. A separate study by Common Sense Media found that over 70% of U.S. teens use AI chatbots for companionship, with half using them regularly.
OpenAI CEO Sam Altman acknowledged the issue of emotional overreliance on AI, noting that some young users feel unable to make decisions without consulting ChatGPT.
Experts say chatbots differ from search engines because they generate personalised responses and are designed to feel human, making them more influential to teens. Younger teens, especially those aged 13 or 14, are more likely to trust chatbot advice.
Common Sense Media has rated ChatGPT as a moderate risk for teens, citing its guardrails. However, the CCDH study shows that these safeguards can be easily bypassed.
Other platforms like Instagram have implemented stronger age verification measures, but ChatGPT has not followed suit.
The study raises concerns about the role of AI in teen mental health and the need for stronger protections to prevent harm.
ChatGPT provided harmful advice to fake teen users, including drug use and self-harm plans
Over 50% of 1,200 chatbot responses were classified as dangerous
ChatGPT does not verify user age or require parental consent
Source: AP NEWS